Objective: We used a variety of health status measures in 2 groups of patients with rheumatoid arthritis (RA) to assess both the smallest distinguishable difference and the relative responsiveness to change of these measures, when used in clinical practice.
Methods: Two groups of patients were studied. Group 1: 24 patients with stable RA tested on 2 occasions; Group 2: 60 patients receiving methotrexate tested before and 14 weeks after treatment with infliximab. Assessments were made with self-completed questionnaires: the modified Health Assessment Questionnaire, Medical Outcomes Study Short Form-36 [SF-36 (SF-6D)], EuroQol, and, in some, the standard gamble. Group 2 also had joint counts, and measures of erythrocyte sedimentation rate, C-reactive protein, and hemoglobin.
Results: The limits-of-agreement (Bland-Altman) approach had greater confidence intervals (CI) than did CI based on +/- 2 standard errors of the measurement. Improvement with infliximab could be determined with all measures, however, but the standard gamble seemed least responsive to change.
Conclusion: The various measures had different degrees of responsiveness, but with all it was possible to show improvement in Group 2 compared to Group 1. There was a closer association of the patient centered measures of improvement with changes in pain score than with joint counts.